Fastening Processing of conversion channel

Hi
I am using ims52 and i have a conversion channel on my mail server.
Now corrospinding to this channel there is a conversion queue.
This queue grows up to a big number and the processing is slow on this channel so the mail reaches to customer in delay.
Please let me know the parameters that i need to modify to fasten the processing on this queue.
Regards
Rajeev Kumar

There was another query, exactly the same as yours, in the last day or two.
Ah there is is. And it is indeed yours. Here's the link to it:
http://swforum.sun.com/jive/thread.jspa?threadID=51178&tstart=15
and what I asked:
If the conversion channel is taking a lot of cpu, it's certanly working.
If you have the original 5.2 bits, and have not applied the p2 patch, there are several, known problems that have been long fixed.
If you have the p2 bits on, you may want to increase the number of processes, or re-examine what you're doing in the conversion channel. Conversion is not as high-performance as much of the rest of the server, as it requires starting a process for each message.
You may simply have more work than the conversion channel can handle on your hardware.
What is it you're doing in conversion?

Similar Messages

Mutiple enqueues to the conversion channel

Hi ,
I did set up a conversion where an AV product
scans an message before sending to the next channel ( basically ims-ms,tcp-local or tcp_intranel ) for delivery.
Recently I noticed that if I send an e-mail to 2 recipients(for instance) , a double copy of the message is enqueued the conversation channel and the AV product scans twice the same message ...
By adding the multiple keywords to the channel option , I can improve things so that if my 2 recipients are to be dequeued to the same channel it is scanned once ( for instance the 2 recipients are to be delivered local delivery)
But if it has to be delivered to 2 different channels ( for instance local and extranet) , it is still scanned twice.
Any option I am missing ? Can I instruct the channel to work only on the envelope of the message ?
Any help appreciated.
Thanks

I see there at least 2 reallly good ideas
Yep, that's how the conversion channel works.First one knowning that Sun suggests to use the
conversion channel for AV purpose , scanning based on
recipients rather than envelope is just a fantastic
for system usage.Our recommendations do change, as we incorporate new features, and our partners add new features, too.
Currently, the conversion channel is the recommended route IF the virus scanner is to be run via command-line. There are indeed some tradeoffs involved with the conversion channel, especially the inherent overhead of starting new processes for each message, address expansion before conversion, and such. This puts a limit on maximum throughput.
>
If your virus scanner works via smtp, you can use
the "aliasdetourhost" keyword instead, and send
such before address expansion. this requires
5.2hf2.02 or later.Second one, if my AV does why should I bother using
iMS , I just use my AV/SMTP package ? While we find that most AV packages are very good at AV, most of them (all I have seen), aren't very good at being a well-behaved SMTP server. The skills that are useful for writing an AV package are very different from those required to create a well-behaved SMTP product.
We find that when most folk expose an AV's smtp port to the general internet, you can get lots of problems, including:
Dropped messages
error messages
much higher load on the AV hardware.
To help deal with such, we recommend using Sun JES' MTA to "wrap" the AV system. By accepting all mail with our MTA, you first get the ability to reject mail that's not addressed to actual users, the ability to throttle connections, to "fix" many types of badly-formed messages, etc.
The new "aliasdetourhost" keyword allows you to select an incoming channel's messages to be sent to an AV system and to accept those messages back into the queue, at high performance, without starting up a new process for each message, and only once per message, even if it's addressed to all of your users. . . .
>
Likely a bad idea, but I have no idea what you did. . .
Thanks for the input even without a clear idea.

Conversion channel and hf 2.07

Hi,
After I applied hf 2.07....I see definite improvement in the way conversion channel is behaving....though I did not do any conversion tuning as suggested by jay...still post hf2.07, the conversion processes have multiplied ...now I can c three conversion processes instead of one....therefore the dequeue also has improved.
I wonder why this has happened ... ! I did not touch the conversion maxjobs or threaddepth!!! just applied hotfix.
Also on my MTA routers ...the conversion processes eventually reduce to one process from 3......even if the conversion queue is having considerable mails in queue....if I restart MTA comp....again three conversion processes start, which dequeue the conversion queued messages fast.....but again when the queue falls to zero ....one conversion process remains....but the processes don't increase as the messages in queueu increases ??
SO I have to restart MTA services for fast conversion dequeue ...!!
Why don't the conversion processes increase in number according to incoming mail traffic ???? Why do multiple conversion processes start only on restarting the MTA (imsimta refresh)????
Also one more thought on conversion tuning....as suggested by jay...threaddepth on conversion ? If i see the refrence guide it states that threaddepth does not apply to conversion and some other channel.....!!!!!
Thanks
MSGADMIN

Hi jay,
I did not add anything on conversion yet.
Just applied the hotfix 2.07.
The performance has definetly improved after hotfix....in a sense...now conversion daemons have multiplied.....and thus the message dequeue in the conversion channel has improved a lot.
But, now that I am observing this new phenomenon, I observerd the conversion processes reduce to one in number...eventually.
When mail traffic goes high...the conversion process does not multiply consistently.....sometimes even 100 mails would not convince the jobcontroller to start multiple conversion processes....and sometimes even 20 mails would do.
A restart would always start 3 conversion processes....
With my settings ....default on conversion ....
default on tcp_local except maxjobs 20
I am having 3 conversion processes ...
I have not done any tuning yet on conversion....

Setting up the Conversion Channel

Hello
I am trying to set up the conversion channel for virus scannin using McAfee products and I cannot get it to work. The mail log shows that it is going through the conversion channel but the scanpm.exe does not execute. Any help would be appreicated.
Chris Toledo
City Colleges of Chicago

Chris, the conversion channel debugs the same way any other channel debugs.
you can bot both master_debug and slave_debug on the channel definition line, recompile the config, and restart the dispatcher
You then get one file per message as the message flows through the system.
Since there are many ways to implement the conversion channel, and the alternate conversion channel, and some of these depend on the exact version of your messaging server (which you neglected to mention), there's really not a lot I can do, yet, to help you.

Infopackage Group to Process Chain Conversion

Dear Gurus,
We are going to do BI upgradation,prrequisite for that we have to convert all Infopackage groups in to Process Chain.In our system totally we 600 Infopackage groups.Let me know the optimized and oraganized way to make a process chain conversion.Before the conversion process, what are the basic measurement (Like sequence,Grouping under process chain,Scheduling .Etc...) I have to do.Anticipate your respone eagerly.
With regards
Siva

Hello Sivakumar,
Try this Function Module RSPC_MIGRATE_INFOPACKAGE_GROUP , this should help you.
You can also goto Process Chain Maintanance in the InfoPackage Group and create the Process Chain.
If that doesn't then as a manual process, you have to note down all the InfoPackage sequence and include them in the Process Chain in the same order.
Thanks
Chandran

Delete specific attachment file through Conversion Channel ?

version : iMS5.2 sp1
O/S : Solaris 2.6 Generic_105181-29
I wanted to delete specific attachment file(ex: ALTDESK.ZIP) through Conversion Channel.
So I set up like below..
1) In mappings file
=================
CONVERSIONS
IN-CHAN=tcp_intranet;OUT-CHAN=tcp_local;CONVERT Yes
==============
I only want to delete attached file from tcp_intranet to tcp_local.
2) msg-INSTANCE/imta/config/conversions
example mail header ::
--- omit ----
MIME-version: 1.0
X-Mailer: iPlanet Messenger Express 5.2 Patch 1 (built Aug 19 2002)
Content-type: multipart/mixed; boundary=--6b2385053506b85
Content-language: ko
X-Accept-Language: ko
Priority: normal
This is a multi-part message in MIME format.
----6b2385053506b85
Content-Type: text/plain; charset=EUC-KR
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable
----6b2385053506b85
Content-Type: application/x-zip-compressed
Content-Transfer-Encoding: base64
Content-Disposition: attachment; filename=ALTDESK.ZIP
- conversions file setting
==================================================
in-channel=tcp_intranet; out-channel=tcp_local;
in-type=application; in-subtype=x-zip-compressed;
parameter-symbol-0=ALTDESK.ZIP; parameter-copy-0=*;
dparameter-symbol-0=ALTDESK.ZIP; dparameter-copy-0=*;
message-header-file=2; original-header-file=1;
override-header-file=1; override-option-file=1;
command="/product/leeky/convert.sh"
============================================
3) /product/leeky/convert.sh file
========================
#!/bin/sh
if [ $? -eq 1 ]; then
echo "STATUS=178030178" >> $OUTPUT_OPTIONS
else
cp $INPUT_FILE $OUTPUT_FILE
fi
=========================
4) The problems I face are
- All of zip files which are filtered is deleted. As you can see 2) I only want to delete ALTDESK.ZIP file
But, all of the zip-compressed files are deleted at the moment.
- This converison channel is work(even if all zip-compressed files are deleted). But sometimes this is not
working(2 or 3 times out of 10). I don not know why.
- Above all, I am not sure 2) and 3) settings are good. English is a second language to me. So it was
not easy to understand conversion channel setting in Admin Guide.
- How can I see the out put of "$OUTPUT_OPTIONS" ? I do not know where I can see that.
Is there anybody to help me ?

The section of the admin guide which can help is:
http://docs.sun.com/source/816-6009-10/channel2.htm#42283
Here there is an explanation of how the mime headers of the message part would align with the entries one would put into the conversions file entry. (The document has an error where it talks about APPARENT_NAME and APPARENT_FILENAME. It should really say the words "NAME" and "FILENAME" respectively).
Based on that document, the MIME headers of your message part :
Content-Type: application/x-zip-compressed
Content-Transfer-Encoding: base64
Content-Disposition: attachment; filename=ALTDESK.ZIP
will align with a conversions file settiong of:
in-channel=tcp_intranet; out-channel=tcp_local;
in-type=application; in-subtype=x-zip-compressed;
parameter-symbol-0=NAME; parameter-copy-0=*;
dparameter-symbol-0=FILENAME; dparameter-copy-0=*;
message-header-file=2; original-header-file=1;
override-header-file=1; override-option-file=1;
command="/product/leeky/convert.sh"
and a /product/leeky/convert.sh script which reads something like:
#!/bin/sh
grep "$FILENAME" /product/leek/badfiles.list
if [ $? -eq 1 ]; then
echo "STATUS=178030178" >> $OUTPUT_OPTIONS
else
grep "$NAME" /product/leek/badfiles.list
if [ $? -eq 1 ]; then
echo "STATUS=178030178" >> $OUTPUT_OPTIONS
else
cp $INPUT_FILE $OUTPUT_FILE
fi
fi
The lines:
parameter-symbol-0=NAME; parameter-copy-0=*;
dparameter-symbol-0=FILENAME; dparameter-copy-0=*;
tells the conversion channel to make the environment variable $NAME avaliable to your program with a value corresponding to the "name=" clause on the Content-Type line of the MIME headers. The environment variable $FILENAME is made avalable to your program and takes on the value extracted from the filename=" clause on the content-disposition line of the MIME headers.
The document at :
http://docs.sun.com/source/816-6092-10/conversion.html
may help provide other examples.

Infopackage groups to process chain conversion

HI Gurus,
I am doing Infopackage Groups to Process chain conversion by using procss chain maintanence icon in scheduler place of Info package group.If I am doing conversion I am getting all infopackages as parallel manner. I need to make it as serial manner.Eagerly expecting your reply.
With regards
Siva

HI shiva,
Can you please clear about ur question:
r u doing infopackage group for ur PC. How many groups u have created.
Generaly if you prepare a batch for infopackages all will run in series( u have to give in proper manner while grouping).
Let me know ur question clearly.
Regards
Ashwin

Could I redirect a message passing by the conversion channel

I Have IMS 5.2 p1, and I am searching about how I can forward a message to a certain mailbox depending on the message attachment. I can only delete the attachment, or hold the message. but cannot redirect the message to another destination recipient ?
Is there a way to do this ?
Please if anyone could help, it is very urgent.
Thanks.

I Have IMS 5.2 p1, and I am searching about how I can
forward a message to a certain mailbox depending on
the message attachment. I can only delete the
attachment, or hold the message. but cannot redirect
the message to another destination recipient ? Actually you probably could redirect this by using
a sieve filter easier than a conversion channel.
But it is possible if you write the appropriate
headers replacing the envelope to: be the new
recipient you could "forward" the message I believe.
For detail support of this nature, I suggest
you call support who can better diagnose
what it is you're trying to do and how best
to accomplish it.
>
Is there a way to do this ?
Please if anyone could help, it is very urgent.
Thanks.

Urgent: Child process closed admin channel on Solaris 9 and 6.1 SP5

Hi
I'm facing this issue on Solaris 9 and Sun AppServer 6.1 SP5, the appserver crashes for some reason and restarts itself. There doesnt seem to be any specific instance which leads to this error. I notice that if I leave my application running for a while I end up seeing this error and I need to kill the instance that was running and restart the server instance.
This is what i see in the log files before the server restarts itself.
" failure (12181): CORE3107: Child process closed admin channel"
It seems like i'm not the only one seeing this issue:
http://swforums.sun.com/jive/thread.jspa?threadID=55040&messageID=210473
http://forum.sun.com/jive/thread.jspa?threadID=95718&messageID=328813
Anybody knows what could be going on?
Thanks

After the corefile is written, use pstack on it. In the pstack file you a stacktrace of the crashing function. Alas, this information is only useful when you have the sources.
Also, usually there are other, more specific lines in the errors logfile e.g.:
[27/Jan/2006:07:38:55] catastrophe (15456): CORE3260: Server crash detected (signal SIGSEGV)
[27/Jan/2006:07:38:55] info (15456): CORE3262: Crash occurred in function INTprepare_nsapi_thread from module /opt/SUNWwbsvr/bin/https/lib/libns-httpd40.so
[27/Jan/2006:07:38:55] failure (15455): CORE3107: Child process closed admin channel
If you post them, maybe somebody can give you a hint...

"CORE3107: Child process closed admin channel" in sun webserver 6.1

Hi,
Does anyone know of the following error in the Sun Java Webserver 6.1 logs:
===========================================
[08/Mar/2010:20:41:12] failure (27868): CORE3107: Child process closed admin channel
[08/Mar/2010:20:41:12] info (13932): HTTP3072: [LS ls1] http://172.21.159.184:80 ready to accept requests
[08/Mar/2010:20:41:12] failure (13932): HTTP3068: Error receiving request from 172.21.158.15 (Connection refused)
[08/Mar/2010:20:41:12] failure (13932): HTTP3068: Error receiving request from 172.21.158.14 (Connection refused)
[08/Mar/2010:20:41:12] failure (13932): HTTP3068: Error receiving request from 172.21.158.15 (Connection refused)
[08/Mar/2010:20:41:12] info (13932): CORE3274: successful server startup
[08/Mar/2010:20:41:12] failure (13932): HTTP3068: Error receiving request from 172.21.158.14 (Connection refused)
[08/Mar/2010:20:45:13] failure (27868): CORE3107: Child process closed admin channel
[08/Mar/2010:20:45:13] info (15075): HTTP3072: [LS ls1] http://172.21.159.184:80 ready to accept requests
[08/Mar/2010:20:45:13] failure (15075): HTTP3068: Error receiving request from 172.21.158.15 (Connection refused)
[08/Mar/2010:20:45:13] failure (15075): HTTP3068: Error receiving request from 172.21.158.14 (Connection refused)
[08/Mar/2010:20:45:13] info (15075): CORE3274: successful server startup
=============================================
We have Sun One Webservers 6.1 installed on Solaris 10 box and where we have installed Siebel Webserver Extension (SWSE) plugin for our siebel application server. Any pointers on how we can fix these crashes(child proc error) on webserver?
Appriciate your help.
Thanks,
Mahipal

Hello were there ever any updates to this post, I'm having basically the same issue and would be interested on an update.

CORE3107: Child process closed admin channel

I Installed SUN ONE Web server 6.1 on a UNIX machine.
Installed 2 web applications.
1 of them works fine (has no Oblix logic), but the other gives me the error message "CORE3107: Child process closed admin channel" and terminates the application.
The point of failure occurs when I call Oblix to initialize.
I have the same web application on IPlanet version 6.0 and has no issues.
Pease help.

That message indicates that the child process terminated unexpectedly. Given your description, the problem is likely a crash in the Oblix plugin.
You may wish to check for core files.
(Note that "UNIX" is not enough information for us to suggest specific steps. Sun ONE Web Server 6.1 is supported on 3 different UNIX platforms, and each has a very different set of diagnostic tools.)

IMS5.2 on W2K Conversion Channel woes

I have been working with the conversion channel in order to scan email attachments for viruses. It all seems to be working with the exception of one thing: the environment variables become intermittantly corrupted. Here is an example:
INPUT_FILE = e:/email/iPlanet/Server5/msg-smtp/imta/tmp/-CO-_ZO7008P0Y4MCCG01RK.00
INPUT_HEADERS = e:/email/iPlanet/Server5/msg-smtp/imta/tmp/-HO-_ZO7008P104MCCG01RK.00��
As you can see, at the end of the value for the INPUT_HEADERS variable there appear to be extended characters. This type of "corruption" appears randomly (in terms of timing as well as as which environment variables are affected).
Due to the length of the path to the different files (INPUT_FILE, INPUT_HEADERS, OUTPUT_OPTIONS, etc.) I can only pass 3 variables via the command line. For the rest of the information I must rely on the environment variables.
How can I get consistant and reliable environment variables?

This is EXACTLY the reason the title page to our forum strongly suggests that you run
imsimta version
and provide that as part of your post.
corruption to conversion files is indeed present in some hotfix versions.
I can confirm if you have such a version, if you would post your exact version.
To get a current/fixed version, you have to open a tech support case.

Sending mails via conversion channel according to sender

Hi
I need to run a script on some mails sent by specific users.
I am using the conversion channel, but in that way all mails using that channel are using the script.
I need that only mails from specific users will use that script.
What is the best method to do so?
Thanks
Shlomi

Hi
I followed the document, but had some problems.
First, When adding the Tag to the mapping file it didn't seem to work, so I put it in the conversion file and then, as it should be, only mails from users that had the tag run via the conversion.
However, it is working only with internal mail.
When sending mails to servers outside my network this tag is ignored.
When sending mail internaly this is the output of the imsimta test -rewrite command:
+# imsimta test -rewrite [email protected] [email protected]+
address channel        =
forward channel        =
backward channel       =
unique identifier      =
header forward address = [email protected], [email protected]
header reverse address = [email protected], [email protected]
envelope forw address =
envelope rev address   =
name                   =
mbox                   =
Extracted address action list:
[email protected]
[email protected]
Extracted 733 address action list:
[email protected]
[email protected]
Address list expansion:
-13 expansion total.
Expanded address:
[email protected], [email protected]
Submitted address list:
ims-ms
+me@ims-ms-daemon (orig [email protected] [email protected], inter [email protected], initial [email protected] [email protected], host ims-ms-daemon) NOTIFY-FAILURES NOTIFY-DELAYS {color:#ff0000}tag 111{color}+
Submitted notifications list:
The tag value is 111, in that case any mail from a user with the value 111 in the mailConversionTag is sent to the conversion channel.
When sending mail to external address I get this:
+# imsimta test -rewrite [email protected] me@external_domain.com+
address channel        =
forward channel        =
backward channel       =
unique identifier      =
header forward address = [email protected], me@external_domain.com
header reverse address = [email protected], me@external_domain.com
envelope forw address =
envelope rev address   =
name                   =
mbox                   =
Extracted address action list:
[email protected]
me@external_domain.com
Extracted 733 address action list:
[email protected]
me@external_domain.com
Address list expansion:
-13 expansion total.
Expanded address:
[email protected], me@external_domain.com
Submitted address list:
ims-ms
+me@ims-ms-daemon (orig [email protected] me@external_domain.com, inter [email protected], initial [email protected] me@external_domain.com, host ims-ms-daemon) NOTIFY-FAILURES NOTIFY-DELAYS {color:#ff0000}tag 111{color}+
tcp_local
+me@external_domain.com (orig [email protected] me@external_domain.com, initial [email protected] me@external_domain.com, host external_domain.com) NOTIFY-FAILURES NOTIFY-DELAYS+
Submitted notifications list:
It looks like after the first phase the tag in not forwarded.
Do I have to enable something so it will be forwarded?
Thanks,
Shlomi

Alternate Conversion Channel

On a previous post, It was recomended that I use
the "Alternate Conversion Channel" setup as per
http://ims.balius.com/downloads/AlternateConversion.pdf
I just seem to have trouble getting it to work. After applying
the settings, The IMS does not change its
behavior at all. Did I miss something?
No messages are going to the tcp_scan channel.
I am using 5.2 Patch 1.
Thanks
Paul

We did that, and the config dat file seems
to be the correct version. Still back to root problem
of "Alternate Conversion Channel" not working as
documented in Chads.
I attached the output of imsimta test.
C:\iPlanet\Server5\msg-icpmail>imsimta test -rewrite [email protected]
Warning - compiled configuration does not match configuration files
-- Modification time mismatch for configuration file C:/iPlanet/Server5/msg-icp
mail/imta/config/option.dat
forward channel = l
channel description =
channel user filter =
dest channel filter =
source channel filter =
channel flags #0 = BIDIRECTIONAL MULTIPLE IMMNONURGENT NOSERVICEALL
channel flags #1 = NOSMTP DEFAULT
channel flags #2 = COPYSENDPOST COPYWARNPOST POSTHEADONLY HEADERINC NOEX
PROUTE
channel flags #3 = NOLOGGING NOGREY NORESTRICTED RETAINSECURITYMULTIPART
S
channel flags #4 = EIGHTBIT NOHEADERTRIM NOHEADERREAD RULES
channel flags #5 =
channel flags #6 = LOCALUSER REPORTHEADER
channel flags #7 = NOSWITCHCHANNEL NOREMOTEHOST DATEFOUR DAYOFWEEK
channel flags #8 = NODEFRAGMENT EXQUOTA REVERSE NOCONVERT_OCTET_STREAM
channel flags #9 = NOTHURMAN INTERPRETENCODING USEINTERMEDIATE RECEIVEDF
ROM VALIDATELOCALSYSTEM NOTURN
defaulthost = icpcorp.com icpcorp.com
linelength = 1023
channel env addr type = SOURCEROUTE
channel hdr addr type = SOURCEROUTE
channel official host = icpmail.icpcorp.com
channel queue 0 name = LOCAL_POOL
channel queue 1 name = LOCAL_POOL
channel queue 2 name = LOCAL_POOL
channel queue 3 name = LOCAL_POOL
channel after params =
channel user name =
urgentnotices = 1 2 4 7
normalnotices = 1 2 4 7
nonurgentnotices = 1 2 4 7
channel rightslist ids =
local behavior flags = %x7
backward channel = l
header To: address = [email protected]
header From: address = [email protected]
envelope To: address = [email protected] (route (icpmail.icpcorp.com,icpma
il.icpcorp.com)) (host icpcorp.com)
envelope From: address = [email protected]
name =
mbox = dtripo
Extracted address action list:
[email protected]
Extracted 733 address action list:
[email protected]
Address list expansion:
dtripo@ims-ms-daemon
1 expansion total.
Expanded address:
[email protected]
Submitted address list:
ims-ms
dtripo@ims-ms-daemon (orig [email protected], host ims-ms-daemon) *NOTIFY-F
AILURES* NOTIFY-DELAYS
Submitted notifications list:
C:\iPlanet\Server5\msg-icpmail>

Conversion channel

Hi jay,
I am planning to replace the conversion channel virus scanning by the aliasdetourhost technique --- which I will be able to implement after hf 2.07.
The conversion channel is actually giving me problems on performance.
So I read ur doc in sunsolve were u have used clamav and amias-new...
I installed it on a test setup and found it interesting ---
The aliasdetourhost lets me to segregate the machines as MTA / STORE and virus/spam scanner...
Just wanted you to suggest me onething ... can i use the sophos in a similar way in tandom with aliasdetourhost thing ...

Sure. Amavis should be able to support that, or if Sophis has it's own smtp wrapper, you can use that.

Fastening Processing of conversion channel

Similar Messages

Maybe you are looking for